No-reference Stereoscopic Image Quality Assessment Using Binocular Self-similarity and Deep Neural Network

作者：

Highlights：

• We propose a new no-reference quality assessment method for stereoscopic images.

• Two indexes, Binocular Self-similarity and binocular integration, are defined.

• We train a Deep Neural Network in an opinion unaware way to predict local quality.

• Experimental results show this method is consistent with subjective perception.

摘要

•We propose a new no-reference quality assessment method for stereoscopic images.•Two indexes, Binocular Self-similarity and binocular integration, are defined.•We train a Deep Neural Network in an opinion unaware way to predict local quality.•Experimental results show this method is consistent with subjective perception.

论文关键词：Stereoscopic image quality assessment,Binocular Self-similarity,Deep Neural Networks,Opinion unaware,Depth image-based rendering

论文评审过程：Received 9 December 2015, Revised 10 July 2016, Accepted 10 July 2016, Available online 12 July 2016, Version of Record 6 August 2016.

论文官网地址：https://doi.org/10.1016/j.image.2016.07.003